builtin/repack.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "config.h"
   4 #include "dir.h"
   5 #include "parse-options.h"
   6 #include "run-command.h"
   7 #include "sigchain.h"
   8 #include "strbuf.h"
   9 #include "string-list.h"
  10 #include "strvec.h"
  11 #include "midx.h"
  12 #include "packfile.h"
  13 #include "prune-packed.h"
  14 #include "object-store.h"
  15 #include "promisor-remote.h"
  16 #include "shallow.h"
  17
  18 static int delta_base_offset = 1;
  19 static int pack_kept_objects = -1;
  20 static int write_bitmaps = -1;
  21 static int use_delta_islands;
  22 static char *packdir, *packtmp;
  23
  24 static const char *const git_repack_usage[] = {
  25         N_("git repack [<options>]"),
  26         NULL
  27 };
  28
  29 static const char incremental_bitmap_conflict_error[] = N_(
  30 "Incremental repacks are incompatible with bitmap indexes.  Use\n"
  31 "--no-write-bitmap-index or disable the pack.writebitmaps configuration."
  32 );
  33
  34
  35 static int repack_config(const char *var, const char *value, void *cb)
  36 {
  37         if (!strcmp(var, "repack.usedeltabaseoffset")) {
  38                 delta_base_offset = git_config_bool(var, value);
  39                 return 0;
  40         }
  41         if (!strcmp(var, "repack.packkeptobjects")) {
  42                 pack_kept_objects = git_config_bool(var, value);
  43                 return 0;
  44         }
  45         if (!strcmp(var, "repack.writebitmaps") ||
  46             !strcmp(var, "pack.writebitmaps")) {
  47                 write_bitmaps = git_config_bool(var, value);
  48                 return 0;
  49         }
  50         if (!strcmp(var, "repack.usedeltaislands")) {
  51                 use_delta_islands = git_config_bool(var, value);
  52                 return 0;
  53         }
  54         return git_default_config(var, value, cb);
  55 }
  56
  57 /*
  58  * Remove temporary $GIT_OBJECT_DIRECTORY/pack/.tmp-$$-pack-* files.
  59  */
  60 static void remove_temporary_files(void)
  61 {
  62         struct strbuf buf = STRBUF_INIT;
  63         size_t dirlen, prefixlen;
  64         DIR *dir;
  65         struct dirent *e;
  66
  67         dir = opendir(packdir);
  68         if (!dir)
  69                 return;
  70
  71         /* Point at the slash at the end of ".../objects/pack/" */
  72         dirlen = strlen(packdir) + 1;
  73         strbuf_addstr(&buf, packtmp);
  74         /* Hold the length of  ".tmp-%d-pack-" */
  75         prefixlen = buf.len - dirlen;
  76
  77         while ((e = readdir(dir))) {
  78                 if (strncmp(e->d_name, buf.buf + dirlen, prefixlen))
  79                         continue;
  80                 strbuf_setlen(&buf, dirlen);
  81                 strbuf_addstr(&buf, e->d_name);
  82                 unlink(buf.buf);
  83         }
  84         closedir(dir);
  85         strbuf_release(&buf);
  86 }
  87
  88 static void remove_pack_on_signal(int signo)
  89 {
  90         remove_temporary_files();
  91         sigchain_pop(signo);
  92         raise(signo);
  93 }
  94
  95 /*
  96  * Adds all packs hex strings to the fname list, which do not
  97  * have a corresponding .keep file. These packs are not to
  98  * be kept if we are going to pack everything into one file.
  99  */
 100 static void get_non_kept_pack_filenames(struct string_list *fname_list,
 101                                         const struct string_list *extra_keep)
 102 {
 103         DIR *dir;
 104         struct dirent *e;
 105         char *fname;
 106
 107         if (!(dir = opendir(packdir)))
 108                 return;
 109
 110         while ((e = readdir(dir)) != NULL) {
 111                 size_t len;
 112                 int i;
 113
 114                 for (i = 0; i < extra_keep->nr; i++)
 115                         if (!fspathcmp(e->d_name, extra_keep->items[i].string))
 116                                 break;
 117                 if (extra_keep->nr > 0 && i < extra_keep->nr)
 118                         continue;
 119
 120                 if (!strip_suffix(e->d_name, ".pack", &len))
 121                         continue;
 122
 123                 fname = xmemdupz(e->d_name, len);
 124
 125                 if (!file_exists(mkpath("%s/%s.keep", packdir, fname)))
 126                         string_list_append_nodup(fname_list, fname);
 127                 else
 128                         free(fname);
 129         }
 130         closedir(dir);
 131 }
 132
 133 static void remove_redundant_pack(const char *dir_name, const char *base_name)
 134 {
 135         struct strbuf buf = STRBUF_INIT;
 136         struct multi_pack_index *m = get_local_multi_pack_index(the_repository);
 137         strbuf_addf(&buf, "%s.pack", base_name);
 138         if (m && midx_contains_pack(m, buf.buf))
 139                 clear_midx_file(the_repository);
 140         strbuf_insertf(&buf, 0, "%s/", dir_name);
 141         unlink_pack_path(buf.buf, 1);
 142         strbuf_release(&buf);
 143 }
 144
 145 struct pack_objects_args {
 146         const char *window;
 147         const char *window_memory;
 148         const char *depth;
 149         const char *threads;
 150         const char *max_pack_size;
 151         int no_reuse_delta;
 152         int no_reuse_object;
 153         int quiet;
 154         int local;
 155 };
 156
 157 static void prepare_pack_objects(struct child_process *cmd,
 158                                  const struct pack_objects_args *args)
 159 {
 160         strvec_push(&cmd->args, "pack-objects");
 161         if (args->window)
 162                 strvec_pushf(&cmd->args, "--window=%s", args->window);
 163         if (args->window_memory)
 164                 strvec_pushf(&cmd->args, "--window-memory=%s", args->window_memory);
 165         if (args->depth)
 166                 strvec_pushf(&cmd->args, "--depth=%s", args->depth);
 167         if (args->threads)
 168                 strvec_pushf(&cmd->args, "--threads=%s", args->threads);
 169         if (args->max_pack_size)
 170                 strvec_pushf(&cmd->args, "--max-pack-size=%s", args->max_pack_size);
 171         if (args->no_reuse_delta)
 172                 strvec_pushf(&cmd->args, "--no-reuse-delta");
 173         if (args->no_reuse_object)
 174                 strvec_pushf(&cmd->args, "--no-reuse-object");
 175         if (args->local)
 176                 strvec_push(&cmd->args,  "--local");
 177         if (args->quiet)
 178                 strvec_push(&cmd->args,  "--quiet");
 179         if (delta_base_offset)
 180                 strvec_push(&cmd->args,  "--delta-base-offset");
 181         strvec_push(&cmd->args, packtmp);
 182         cmd->git_cmd = 1;
 183         cmd->out = -1;
 184 }
 185
 186 /*
 187  * Write oid to the given struct child_process's stdin, starting it first if
 188  * necessary.
 189  */
 190 static int write_oid(const struct object_id *oid, struct packed_git *pack,
 191                      uint32_t pos, void *data)
 192 {
 193         struct child_process *cmd = data;
 194
 195         if (cmd->in == -1) {
 196                 if (start_command(cmd))
 197                         die(_("could not start pack-objects to repack promisor objects"));
 198         }
 199
 200         xwrite(cmd->in, oid_to_hex(oid), the_hash_algo->hexsz);
 201         xwrite(cmd->in, "\n", 1);
 202         return 0;
 203 }
 204
 205 static struct {
 206         const char *name;
 207         unsigned optional:1;
 208 } exts[] = {
 209         {".pack"},
 210         {".idx"},
 211         {".bitmap", 1},
 212         {".promisor", 1},
 213 };
 214
 215 static unsigned populate_pack_exts(char *name)
 216 {
 217         struct stat statbuf;
 218         struct strbuf path = STRBUF_INIT;
 219         unsigned ret = 0;
 220         int i;
 221
 222         for (i = 0; i < ARRAY_SIZE(exts); i++) {
 223                 strbuf_reset(&path);
 224                 strbuf_addf(&path, "%s-%s%s", packtmp, name, exts[i].name);
 225
 226                 if (stat(path.buf, &statbuf))
 227                         continue;
 228
 229                 ret |= (1 << i);
 230         }
 231
 232         strbuf_release(&path);
 233         return ret;
 234 }
 235
 236 static void repack_promisor_objects(const struct pack_objects_args *args,
 237                                     struct string_list *names)
 238 {
 239         struct child_process cmd = CHILD_PROCESS_INIT;
 240         FILE *out;
 241         struct strbuf line = STRBUF_INIT;
 242
 243         prepare_pack_objects(&cmd, args);
 244         cmd.in = -1;
 245
 246         /*
 247          * NEEDSWORK: Giving pack-objects only the OIDs without any ordering
 248          * hints may result in suboptimal deltas in the resulting pack. See if
 249          * the OIDs can be sent with fake paths such that pack-objects can use a
 250          * {type -> existing pack order} ordering when computing deltas instead
 251          * of a {type -> size} ordering, which may produce better deltas.
 252          */
 253         for_each_packed_object(write_oid, &cmd,
 254                                FOR_EACH_OBJECT_PROMISOR_ONLY);
 255
 256         if (cmd.in == -1)
 257                 /* No packed objects; cmd was never started */
 258                 return;
 259
 260         close(cmd.in);
 261
 262         out = xfdopen(cmd.out, "r");
 263         while (strbuf_getline_lf(&line, out) != EOF) {
 264                 struct string_list_item *item;
 265                 char *promisor_name;
 266                 int fd;
 267                 if (line.len != the_hash_algo->hexsz)
 268                         die(_("repack: Expecting full hex object ID lines only from pack-objects."));
 269                 item = string_list_append(names, line.buf);
 270
 271                 /*
 272                  * pack-objects creates the .pack and .idx files, but not the
 273                  * .promisor file. Create the .promisor file, which is empty.
 274                  *
 275                  * NEEDSWORK: fetch-pack sometimes generates non-empty
 276                  * .promisor files containing the ref names and associated
 277                  * hashes at the point of generation of the corresponding
 278                  * packfile, but this would not preserve their contents. Maybe
 279                  * concatenate the contents of all .promisor files instead of
 280                  * just creating a new empty file.
 281                  */
 282                 promisor_name = mkpathdup("%s-%s.promisor", packtmp,
 283                                           line.buf);
 284                 fd = open(promisor_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 285                 if (fd < 0)
 286                         die_errno(_("unable to create '%s'"), promisor_name);
 287                 close(fd);
 288
 289                 item->util = (void *)(uintptr_t)populate_pack_exts(item->string);
 290
 291                 free(promisor_name);
 292         }
 293         fclose(out);
 294         if (finish_command(&cmd))
 295                 die(_("could not finish pack-objects to repack promisor objects"));
 296 }
 297
 298 #define ALL_INTO_ONE 1
 299 #define LOOSEN_UNREACHABLE 2
 300
 301 int cmd_repack(int argc, const char **argv, const char *prefix)
 302 {
 303         struct child_process cmd = CHILD_PROCESS_INIT;
 304         struct string_list_item *item;
 305         struct string_list names = STRING_LIST_INIT_DUP;
 306         struct string_list rollback = STRING_LIST_INIT_NODUP;
 307         struct string_list existing_packs = STRING_LIST_INIT_DUP;
 308         struct strbuf line = STRBUF_INIT;
 309         int i, ext, ret;
 310         FILE *out;
 311
 312         /* variables to be filled by option parsing */
 313         int pack_everything = 0;
 314         int delete_redundant = 0;
 315         const char *unpack_unreachable = NULL;
 316         int keep_unreachable = 0;
 317         struct string_list keep_pack_list = STRING_LIST_INIT_NODUP;
 318         int no_update_server_info = 0;
 319         struct pack_objects_args po_args = {NULL};
 320
 321         struct option builtin_repack_options[] = {
 322                 OPT_BIT('a', NULL, &pack_everything,
 323                                 N_("pack everything in a single pack"), ALL_INTO_ONE),
 324                 OPT_BIT('A', NULL, &pack_everything,
 325                                 N_("same as -a, and turn unreachable objects loose"),
 326                                    LOOSEN_UNREACHABLE | ALL_INTO_ONE),
 327                 OPT_BOOL('d', NULL, &delete_redundant,
 328                                 N_("remove redundant packs, and run git-prune-packed")),
 329                 OPT_BOOL('f', NULL, &po_args.no_reuse_delta,
 330                                 N_("pass --no-reuse-delta to git-pack-objects")),
 331                 OPT_BOOL('F', NULL, &po_args.no_reuse_object,
 332                                 N_("pass --no-reuse-object to git-pack-objects")),
 333                 OPT_BOOL('n', NULL, &no_update_server_info,
 334                                 N_("do not run git-update-server-info")),
 335                 OPT__QUIET(&po_args.quiet, N_("be quiet")),
 336                 OPT_BOOL('l', "local", &po_args.local,
 337                                 N_("pass --local to git-pack-objects")),
 338                 OPT_BOOL('b', "write-bitmap-index", &write_bitmaps,
 339                                 N_("write bitmap index")),
 340                 OPT_BOOL('i', "delta-islands", &use_delta_islands,
 341                                 N_("pass --delta-islands to git-pack-objects")),
 342                 OPT_STRING(0, "unpack-unreachable", &unpack_unreachable, N_("approxidate"),
 343                                 N_("with -A, do not loosen objects older than this")),
 344                 OPT_BOOL('k', "keep-unreachable", &keep_unreachable,
 345                                 N_("with -a, repack unreachable objects")),
 346                 OPT_STRING(0, "window", &po_args.window, N_("n"),
 347                                 N_("size of the window used for delta compression")),
 348                 OPT_STRING(0, "window-memory", &po_args.window_memory, N_("bytes"),
 349                                 N_("same as the above, but limit memory size instead of entries count")),
 350                 OPT_STRING(0, "depth", &po_args.depth, N_("n"),
 351                                 N_("limits the maximum delta depth")),
 352                 OPT_STRING(0, "threads", &po_args.threads, N_("n"),
 353                                 N_("limits the maximum number of threads")),
 354                 OPT_STRING(0, "max-pack-size", &po_args.max_pack_size, N_("bytes"),
 355                                 N_("maximum size of each packfile")),
 356                 OPT_BOOL(0, "pack-kept-objects", &pack_kept_objects,
 357                                 N_("repack objects in packs marked with .keep")),
 358                 OPT_STRING_LIST(0, "keep-pack", &keep_pack_list, N_("name"),
 359                                 N_("do not repack this pack")),
 360                 OPT_END()
 361         };
 362
 363         git_config(repack_config, NULL);
 364
 365         argc = parse_options(argc, argv, prefix, builtin_repack_options,
 366                                 git_repack_usage, 0);
 367
 368         if (delete_redundant && repository_format_precious_objects)
 369                 die(_("cannot delete packs in a precious-objects repo"));
 370
 371         if (keep_unreachable &&
 372             (unpack_unreachable || (pack_everything & LOOSEN_UNREACHABLE)))
 373                 die(_("--keep-unreachable and -A are incompatible"));
 374
 375         if (write_bitmaps < 0) {
 376                 if (!(pack_everything & ALL_INTO_ONE) ||
 377                     !is_bare_repository())
 378                         write_bitmaps = 0;
 379         }
 380         if (pack_kept_objects < 0)
 381                 pack_kept_objects = write_bitmaps > 0;
 382
 383         if (write_bitmaps && !(pack_everything & ALL_INTO_ONE))
 384                 die(_(incremental_bitmap_conflict_error));
 385
 386         packdir = mkpathdup("%s/pack", get_object_directory());
 387         packtmp = mkpathdup("%s/.tmp-%d-pack", packdir, (int)getpid());
 388
 389         sigchain_push_common(remove_pack_on_signal);
 390
 391         prepare_pack_objects(&cmd, &po_args);
 392
 393         strvec_push(&cmd.args, "--keep-true-parents");
 394         if (!pack_kept_objects)
 395                 strvec_push(&cmd.args, "--honor-pack-keep");
 396         for (i = 0; i < keep_pack_list.nr; i++)
 397                 strvec_pushf(&cmd.args, "--keep-pack=%s",
 398                              keep_pack_list.items[i].string);
 399         strvec_push(&cmd.args, "--non-empty");
 400         strvec_push(&cmd.args, "--all");
 401         strvec_push(&cmd.args, "--reflog");
 402         strvec_push(&cmd.args, "--indexed-objects");
 403         if (has_promisor_remote())
 404                 strvec_push(&cmd.args, "--exclude-promisor-objects");
 405         if (write_bitmaps > 0)
 406                 strvec_push(&cmd.args, "--write-bitmap-index");
 407         else if (write_bitmaps < 0)
 408                 strvec_push(&cmd.args, "--write-bitmap-index-quiet");
 409         if (use_delta_islands)
 410                 strvec_push(&cmd.args, "--delta-islands");
 411
 412         if (pack_everything & ALL_INTO_ONE) {
 413                 get_non_kept_pack_filenames(&existing_packs, &keep_pack_list);
 414
 415                 repack_promisor_objects(&po_args, &names);
 416
 417                 if (existing_packs.nr && delete_redundant) {
 418                         if (unpack_unreachable) {
 419                                 strvec_pushf(&cmd.args,
 420                                              "--unpack-unreachable=%s",
 421                                              unpack_unreachable);
 422                                 strvec_push(&cmd.env_array, "GIT_REF_PARANOIA=1");
 423                         } else if (pack_everything & LOOSEN_UNREACHABLE) {
 424                                 strvec_push(&cmd.args,
 425                                             "--unpack-unreachable");
 426                         } else if (keep_unreachable) {
 427                                 strvec_push(&cmd.args, "--keep-unreachable");
 428                                 strvec_push(&cmd.args, "--pack-loose-unreachable");
 429                         } else {
 430                                 strvec_push(&cmd.env_array, "GIT_REF_PARANOIA=1");
 431                         }
 432                 }
 433         } else {
 434                 strvec_push(&cmd.args, "--unpacked");
 435                 strvec_push(&cmd.args, "--incremental");
 436         }
 437
 438         cmd.no_stdin = 1;
 439
 440         ret = start_command(&cmd);
 441         if (ret)
 442                 return ret;
 443
 444         out = xfdopen(cmd.out, "r");
 445         while (strbuf_getline_lf(&line, out) != EOF) {
 446                 if (line.len != the_hash_algo->hexsz)
 447                         die(_("repack: Expecting full hex object ID lines only from pack-objects."));
 448                 string_list_append(&names, line.buf);
 449         }
 450         fclose(out);
 451         ret = finish_command(&cmd);
 452         if (ret)
 453                 return ret;
 454
 455         if (!names.nr && !po_args.quiet)
 456                 printf_ln(_("Nothing new to pack."));
 457
 458         for_each_string_list_item(item, &names) {
 459                 item->util = (void *)(uintptr_t)populate_pack_exts(item->string);
 460         }
 461
 462         close_object_store(the_repository->objects);
 463
 464         /*
 465          * Ok we have prepared all new packfiles.
 466          */
 467         for_each_string_list_item(item, &names) {
 468                 for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
 469                         char *fname, *fname_old;
 470
 471                         fname = mkpathdup("%s/pack-%s%s",
 472                                         packdir, item->string, exts[ext].name);
 473                         fname_old = mkpathdup("%s-%s%s",
 474                                         packtmp, item->string, exts[ext].name);
 475
 476                         if (((uintptr_t)item->util) & (1 << ext)) {
 477                                 struct stat statbuffer;
 478                                 if (!stat(fname_old, &statbuffer)) {
 479                                         statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
 480                                         chmod(fname_old, statbuffer.st_mode);
 481                                 }
 482
 483                                 if (rename(fname_old, fname))
 484                                         die_errno(_("renaming '%s' failed"), fname_old);
 485                         } else if (!exts[ext].optional)
 486                                 die(_("missing required file: %s"), fname_old);
 487                         else if (unlink(fname) < 0 && errno != ENOENT)
 488                                 die_errno(_("could not unlink: %s"), fname);
 489
 490                         free(fname);
 491                         free(fname_old);
 492                 }
 493         }
 494         /* End of pack replacement. */
 495
 496         reprepare_packed_git(the_repository);
 497
 498         if (delete_redundant) {
 499                 const int hexsz = the_hash_algo->hexsz;
 500                 int opts = 0;
 501                 string_list_sort(&names);
 502                 for_each_string_list_item(item, &existing_packs) {
 503                         char *sha1;
 504                         size_t len = strlen(item->string);
 505                         if (len < hexsz)
 506                                 continue;
 507                         sha1 = item->string + len - hexsz;
 508                         if (!string_list_has_string(&names, sha1))
 509                                 remove_redundant_pack(packdir, item->string);
 510                 }
 511                 if (!po_args.quiet && isatty(2))
 512                         opts |= PRUNE_PACKED_VERBOSE;
 513                 prune_packed_objects(opts);
 514
 515                 if (!keep_unreachable &&
 516                     (!(pack_everything & LOOSEN_UNREACHABLE) ||
 517                      unpack_unreachable) &&
 518                     is_repository_shallow(the_repository))
 519                         prune_shallow(PRUNE_QUICK);
 520         }
 521
 522         if (!no_update_server_info)
 523                 update_server_info(0);
 524         remove_temporary_files();
 525
 526         if (git_env_bool(GIT_TEST_MULTI_PACK_INDEX, 0))
 527                 write_midx_file(get_object_directory(), 0);
 528
 529         string_list_clear(&names, 0);
 530         string_list_clear(&rollback, 0);
 531         string_list_clear(&existing_packs, 0);
 532         strbuf_release(&line);
 533
 534         return 0;
 535 }